我一直在尝试制作一个NN，该NN将评估来自树莓派上流的2秒数据块。训练了NN，并控制了流，但是为了减少延迟，我们希望使用线程不断地对2秒块进行评估。
我一直在这里使用python（https://realpython.com/intro-to-python-threading/）示例的线程包。
这是模型：
drop_out_rate = 0.1
learning_rate = 0.001
number_of_epochs = 100
number_of_classes = 2
batch_size = 32
optimizer = optimizers.Adam(learning_rate, learning_rate / 100)
input_tensor = Input(shape=input_shape)
metrics = [auc, "accuracy"]
x = layers.Conv1D(16, 9, activation="relu", padding="same")(input_tensor)
x = layers.Conv1D(16, 9, activation="relu", padding="same")(x)
x = layers.MaxPool1D(16)(x)
x = layers.Dropout(rate=drop_out_rate)(x)
x = layers.Conv1D(32, 3, activation="relu", padding="same")(x)
x = layers.Conv1D(32, 3, activation="relu", padding="same")(x)
x = layers.MaxPool1D(4)(x)
x = layers.Dropout(rate=drop_out_rate)(x)
x = layers.Conv1D(32, 3, activation="relu", padding="same")(x)
x = layers.Conv1D(32, 3, activation="relu", padding="same")(x)
x = layers.MaxPool1D(4)(x)
x = layers.Dropout(rate=drop_out_rate)(x)
x = layers.Conv1D(256, 3, activation="relu", padding="same")(x)
x = layers.Conv1D(256, 3, activation="relu", padding="same")(x)
x = layers.GlobalMaxPool1D()(x)
x = layers.Dropout(rate=(drop_out_rate * 2))(x) # Increasing drop-out rate here to prevent overfitting
x = layers.Dense(64, activation="relu")(x)
x = layers.Dense(1028, activation="relu")(x)
output_tensor = layers.Dense(number_of_classes, activation="softmax")(x)
model = tf.keras.Model(input_tensor, output_tensor)
model.compile(optimizer=optimizer, loss=keras.losses.binary_crossentropy, metrics=metrics)
model.load_weights(os.getcwd()+"/gunshot_detection/raspberry_pi/models/gunshot_sound_model.h5")
这是线程功能：
def thread_function(microphone_data,name):
tf.keras.backend.clear_session()
logging.info("Thread %s: starting", name)
reformed_microphone_data = librosa.util.normalize(microphone_data)
reformed_microphone_data = reformed_microphone_data.reshape(-1, audio_rate, 1)
print(reformed_microphone_data.shape, "this")
# Passes a given audio sample into the model for prediction
probabilities = model.predict(reformed_microphone_data)
print(probabilities)
logging.info("Probability of %s: %s", name,str(probabilities))
logging.info("Thread %s: finishing", name)
这就是它们的名称：
format = "%(asctime)s: %(message)s"
logging.basicConfig(format=format, level=logging.INFO,
datefmt="%H:%M:%S")
np_arrays = np.load(path)
threads = list()
for index in range(5):
logging.info("Main : create and start thread %d.", index)
a = np_arrays[index].reshape(input_shape)
#print(index,a.shape)
x = threading.Thread(target=thread_function, args=(a,index,))
threads.append(x)
x.start()
for index, thread in enumerate(threads):
logging.info("Main : before joining thread %d.", index)
thread.join()
logging.info("Main : thread %d done", index)
这样做时，我们在模型的预测调用中遇到了这个错误：
Exception in thread Thread-101:
Traceback (most recent call last):
File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/threading.py", line 917, in _bootstrap_inner
self.run()
File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/threading.py", line 865, in run
self._target(*self._args, **self._kwargs)
File "<ipython-input-138-e8e46f092078>", line 11, in thread_function
probabilities = model.predict(reformed_microphone_data)
File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/tensorflow/python/keras/engine/training.py", line 1113, in predict
self, x, batch_size=batch_size, verbose=verbose, steps=steps)
File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/tensorflow/python/keras/engine/training_arrays.py", line 195, in model_iteration
f = _make_execution_function(model, mode)
File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/tensorflow/python/keras/engine/training_arrays.py", line 122, in _make_execution_function
return model._make_execution_function(mode)
File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/tensorflow/python/keras/engine/training.py", line 1989, in _make_execution_function
self._make_predict_function()
File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/tensorflow/python/keras/engine/training.py", line 1979, in _make_predict_function
**kwargs)
File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/tensorflow/python/keras/backend.py", line 3201, in function
return GraphExecutionFunction(inputs, outputs, updates=updates, **kwargs)
File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/tensorflow/python/keras/backend.py", line 2939, in __init__
with ops.control_dependencies(self.outputs):
File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/tensorflow/python/framework/ops.py", line 5028, in control_dependencies
return get_default_graph().control_dependencies(control_inputs)
File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/tensorflow/python/framework/ops.py", line 4528, in control_dependencies
c = self.as_graph_element(c)
File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/tensorflow/python/framework/ops.py", line 3478, in as_graph_element
return self._as_graph_element_locked(obj, allow_tensor, allow_operation)
File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/tensorflow/python/framework/ops.py", line 3557, in _as_graph_element_locked
raise ValueError("Tensor %s is not an element of this graph." % obj)
ValueError: Tensor Tensor("dense_17/Softmax:0", shape=(?, 2), dtype=float32) is not an element of this graph.
这是一个失败的原因吗？在对音频进行评估时，有什么方法可以同时不断地流音频？如果您有其他解决方案，请告诉我。